Approximation error analysis of some deep backward schemes for nonlinear PDEs
From MaRDI portal
Publication:5021399
backward SDEsdeep splittingerror estimatesmultistep schemesneural networksnonlinear PDEsnumerical approximation
Computational methods for stochastic equations (aspects of stochastic analysis) (60H35) Probabilistic models, generic numerical methods in probability and statistics (65C20) Stability and convergence of numerical methods for initial value and initial-boundary value problems involving PDEs (65M12) Numerical methods for partial differential equations, initial value and time-dependent initial-boundary value problems (65M99)
Abstract: Recently proposed numerical algorithms for solving high-dimensional nonlinear partial differential equations (PDEs) based on neural networks have shown their remarkable performance. We review some of them and study their convergence properties. The methods rely on probabilistic representation of PDEs by backward stochastic differential equations (BSDEs) and their iterated time discretization. Our proposed algorithm, called deep backward multistep scheme (MDBDP), is a machine learning version of the LSMDP scheme of Gobet, Turkedjiev (Math. Comp. 85, 2016). It estimates simultaneously by backward induction the solution and its gradient by neural networks through sequential minimizations of suitable quadratic loss functions that are performed by stochastic gradient descent. Our main theoretical contribution is to provide an approximation error analysis of the MDBDP scheme as well as the deep splitting (DS) scheme for semilinear PDEs designed in Beck, Becker, Cheridito, Jentzen, Neufeld (2019). We also supplement the error analysis of the DBDP scheme of Hur{'e}, Pham, Warin (Math. Comp. 89, 2020). This yields notably convergence rate in terms of the number of neurons for a class of deep Lipschitz continuous GroupSort neural networks when the PDE is linear in the gradient of the solution for the MDBDP scheme, and in the semilinear case for the DBDP scheme. We illustrate our results with some numerical tests that are compared with some other machine learning algorithms in the literature.
Recommendations
- Deep backward schemes for high-dimensional nonlinear PDEs
- Deep learning-based numerical methods for high-dimensional parabolic partial differential equations and backward stochastic differential equations
- Machine learning approximation algorithms for high-dimensional fully nonlinear partial differential equations and second-order backward stochastic differential equations
- Algorithms for solving high dimensional PDEs: from nonlinear Monte Carlo to machine learning
- Control variate method for deep BSDE solver using weak approximation
Cites work
- A distribution-free theory of nonparametric regression
- A forward scheme for backward SDEs
- A numerical scheme for BSDEs
- A probabilistic numerical method for fully nonlinear parabolic PDEs
- A regression-based Monte Carlo method to solve backward stochastic differential equations
- Adapted solution of a backward stochastic differential equation
- Backward stochastic differential equations. From linear to fully nonlinear theory
- Branching diffusion representation of semilinear PDEs and Monte Carlo approximation
- Breaking the curse of dimensionality with convex neural networks
- Convergence of the deep BSDE method for coupled FBSDEs
- Deep backward schemes for high-dimensional nonlinear PDEs
- Deep neural networks algorithms for stochastic control problems on finite horizon: convergence analysis
- Deep splitting method for parabolic PDEs
- DGM: a deep learning algorithm for solving partial differential equations
- Discrete-time approximation and Monte-Carlo simulation of backward stochastic differential equations
- Error bounds for approximations with deep ReLU networks
- scientific article; zbMATH DE number 1405266 (Why is no real title available?)
- Linear regression MDP scheme for discrete backward stochastic differential equations under general conditions
- Machine learning for semi linear PDEs
- Multilayer feedforward networks are universal approximators
- On multilevel Picard numerical approximations for high-dimensional nonlinear parabolic partial differential equations and high-dimensional nonlinear backward stochastic differential equations
- Solving high-dimensional Hamilton-Jacobi-Bellman PDEs using neural networks: perspectives from the theory of controlled diffusions and measures on path space
- Solving high-dimensional partial differential equations using deep learning
Cited in
(39)- Neural networks-based backward scheme for fully nonlinear PDEs
- Gradient convergence of deep learning-based numerical methods for BSDEs
- Gradient boosting-based numerical methods for high-dimensional backward stochastic differential equations
- DeepSets and their derivative networks for solving symmetric PDEs
- Convergence of deep fictitious play for stochastic differential games
- An overview on deep learning-based approximation methods for partial differential equations
- Deep backward schemes for high-dimensional nonlinear PDEs
- A Neural Network Approach to High-Dimensional Optimal Switching Problems with Jumps in Energy Markets
- Neural networks for first order HJB equations and application to front propagation with obstacle terms
- The Effect of the Number of Neural Networks on Deep Learning Schemes for Solving High Dimensional Nonlinear Backward Stochastic Differential Equations
- Numerical methods for backward stochastic differential equations: a survey
- Solving Kolmogorov PDEs without the curse of dimensionality via deep learning and asymptotic expansion with Malliavin calculus
- Deep learning algorithms for solving high-dimensional nonlinear backward stochastic differential equations
- A gradient method for high-dimensional BSDEs
- A deep learning method for solving multi-dimensional coupled forward-backward doubly SDEs
- High-dimensional stochastic control models for newsvendor problems and deep learning resolution
- Numerical approximation of Dynkin games with asymmetric information
- Generalization error analysis of deep backward dynamic programming for solving nonlinear PDEs
- An entropy penalized approach for stochastic control problems
- Deep learning for solving initial path optimization of mean-field systems with memory
- Deep Picard iteration for high-dimensional nonlinear PDEs
- Multilevel Picard approximations overcome the curse of dimensionality when approximating semilinear heat equations with gradient-dependent nonlinearities in L^p-sense
- Mean-field neural networks-based algorithms for McKean-Vlasov control problems
- SOC-MartNet: a martingale neural network for the Hamilton-Jacobi-Bellman equation without explicit \(\inf_{u\in U}H\) in stochastic optimal controls
- Multilevel Picard approximations overcome the curse of dimensionality in the numerical approximation of general semilinear PDEs with gradient-dependent nonlinearities
- Deep learning numerical methods for high-dimensional quasilinear PIDEs and coupled FBSDEs with jumps
- Full error analysis of the random deep splitting method for nonlinear parabolic PDEs and PIDEs
- Convergence analysis of the deep splitting scheme: the case of partial integro-differential equations and the associated forward backward SDEs with jumps
- A gradient method for fully nonlinear parabolic differential equations with convex nonlinearity
- Local discontinuous Galerkin method for nonlinear BSPDEs of Neumann boundary conditions with deep backward dynamic programming time-marching
- A deep solver for backward stochastic Volterra integral equations
- A backward differential deep learning-based algorithm for solving high-dimensional nonlinear backward stochastic differential equations
- Deep random difference method for high-dimensional quasilinear parabolic partial differential equations
- Multilevel Picard algorithm for general semilinear parabolic PDEs with gradient-dependent nonlinearities
- Multilevel Picard approximation algorithm for semilinear partial integro-differential equations and its complexity analysis
- Convergence analysis of iterative deep learning algorithms for fully nonlinear BSPDEs in non-Markovian utility maximization
- XGBoost Monte Carlo high-order regression method for high-dimensional decoupled forward backward stochastic differential equations
- Multilevel Picard approximations for McKean-Vlasov stochastic differential equations with jumps
- A deep BSDE approach for the simultaneous pricing and delta-gamma hedging of large portfolios consisting of high-dimensional multi-asset Bermudan options
This page was built for publication: Approximation error analysis of some deep backward schemes for nonlinear PDEs
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5021399)
